AI as a Senior Engineer's Apprentice

There’s a familiar pattern in how engineers talk about AI coding tools. They either dismiss them as glorified autocomplete or overclaim they’re about to replace the entire profession. Both framings miss the point.

After two decades in the industry, the mental model I’ve found most useful is simpler: AI is an apprentice. A capable one.

An apprentice who can execute a task you describe clearly. Who learns from the examples you give it. Who works fast and rarely complains. And who, like any apprentice, needs clear direction and careful review.

The senior part matters

What makes this framing useful is what it implies about the other side of the equation. Apprentices are only as productive as the seniors guiding them. You can’t hand someone a vague requirement and expect clean output. You need to know what you want, how to describe it, and how to recognize when it’s wrong.

Those aren’t skills AI gives you. They’re skills you bring to it.

I find AI most useful when I already understand the domain deeply enough to ask precise questions. When I know the shape of what I want, it helps me get there faster. When I’m confused about the fundamentals, it generates plausible-sounding confusion back at me.

What this means in practice

I use AI for things I’d have handed to a junior engineer ten years ago: first drafts of boilerplate, test generation, documentation, translating an idea into working syntax in an unfamiliar language. I review everything. I catch things. I push back.

What I don’t use it for is thinking. It can’t tell me whether I’m solving the right problem. It can’t push back on a product assumption. It can’t recognize when a simple answer masks a systemic issue.

That’s still my job. The judgment is still mine.

The real shift

The shift AI creates isn’t about speed. It’s about leverage. A senior engineer with good AI tooling can now cover more ground — prototype faster, produce more, maintain more surface area — without sacrificing the thinking that makes the output worth having.

That’s a meaningful change. But it only works if the senior part stays intact.